Proc. Nat. Acad. Sci. USAVol. 73, No. 3, pp. 742-746, March 1976Biochemistry

Concatemers of alternating plus and minus strands areintermediates in adenovirus-associated virus DNA synthesis*

(DNA replication/hairpin molecules/self-primed DNA synthesis)

STEPHEN E. STRAUS, EDWIN D. SEBRING, AND JAMES A. ROSELaboratory of Biology of Viruses, National Institute of Allergy and Infectious Diseases, NIH, Bethesda, Maryland 20014

Communicated by Wallace P. Rowe, December 9, 1975

ABSTRACT Replicating DNA molecules of adenovirus-associated virus (AAV) were selectively extracted' from KBcells coinfected at 39.5° with a DNA minus, temperature-sen-sitive mutant of adenovirus 5 (ts125) as helper. Under theseconditions AAV DNA replication proceeds normally, butthere is little, if any, adenovirus DNA synthesis. An analysisof the replicating molecules in sucrose density gradients re-veals that there are AAV DNA intermediates which consistof covalently linked plus and minus DNA strands. Under de-naturing conditions, these concatemers are linear singlestrands whose lengths can reach at least four times the sizeof the AAV genome. The most abundant concatemericspecies is a dimer which presumably exists in vivo as a unitlength hairpin. Unit length linear duplexes appear to be im-mediate precursors of plus and minus progeny strands. Thesefindings are compatible with a self-priming mechanism forthe synthesis of AAV DNA.

Adenovirus-associated viruses (AAV) contain single-strand-ed, linear DNA with a molecular weight of about 1.4 X 106(1). They are defective parvoviruses that multiply only incells coinfected with a helper adenovirus that may provideone or more factors required for AAV DNA synthesis (2).There are two unusual genetic features of the AAV: (a) plusor minus DNA strands are separately packaged in progenyvirions and (b) both of these complementary strands possessan inverted terminal repetition- 100-200 nucleotides inlength (3-6). The function of the repetitious sequences is notdefinitely known, but their location and the possibility thatthey contain a terminal palindrome suggest a role in DNAsynthesis (5, 7). Furthermore, this repetition may relate toadenovirus dependence, since the adenovirus genome is theonly other viral DNA known to possess a similar type of ter-minal redundancy (8-10).The present study was undertaken to characterize AAV

DNA replicative intermediates (RI). Our work was facilitat-ed by using a DNA minus, temperature-sensitive mutant ofadenovirus 5 (ts125) as helper. Ts125 supports normal AAVmultiplication at the restrictive temperature (39.50), butthere is little or no detectable adenovirus DNA synthesis(11). Thus, when selectively extracted by a modification ofthe Hirt procedure, replicating components of AAV DNAare essentially free of contaminating adenovirus DNA (11).In this report we present a detailed analysis of such extractsin neutral and alkaline sucrose density gradients. The resultsprovide evidence for concatemeric DNA intermediateswhich suggest that AAV DNA replication proceeds by a

Abbreviations: AAV, adenovirus-associated virus; Ad, adenovirus;RI, replicative intermediates; TCID50, 50% tissue culture infectivedose; WT, wild type; SV40, simian virus 40; BND-cellulose, benzo-ylated-naphtholated DEAE-cellulose; DS, double stranded fraction.* Presented in part at the meeting of Federation of American So-cieties for Experimental Biology, April 1975 [(1975) Fed. Proc.34, 639].

unique mechanism in which DNA synthesis is initiated by aself-priming terminal sequence.

MATERIALS AND METHODSCells and Viruses. Infections were carried out in KB cell

spinner cultures at 39.5'. Both adenovirus 5 (AdS) wild type(WT) and ts125 were a generous gift of H. S. Ginsberg. Pro-cedures for the growth, purification, and assay of thesestrains as well as for AAV type 2 (AAV-2) have been de-scribed (11, 12). The Ad5 inocula were cell-free concentrates(12); the AAV-2 inoculum consisted of heated, CsCI-purifiedvirus (11). Ad-AAV coinfection multiplicities were 5 TCID50and 2 TCID5o units per cell. (TCID50 is the 50% tissue cul-ture infective dose.)

Labeling and Extraction of DNA. AAV-2 DNA waspulse-labeled with [3H]thymidine (50 Ci/mmol) added tocultures (10 0Ci/ml) at 16 hr after infection for intervalsspecified in individual experiments. [3H]Thymidine chaseswere performed by pelleting pulse-labeled, infected cells at600 X g and resuspending cells in warmed growth mediumsupplemented with thymidine, 100 ,4g/ml, and 2'-deoxycy-tidine, 10 ,ug/ml. These conditions permit continued DNAsynthesis without further incorporation of [3H]thymidine(13). Methods for extracting whole cell DNA and DNA frompurified virus preparations have been published (14). Theselective recovery of viral DNA from infected cells was ac-complished by a modification of the Hirt procedure (11).DNA was extracted from pellet fractions as described forwhole cell DNA except that pellets were first heated to 80°in 0.5 M NaOH and then neutralized.

Analysis and Recovery of DNA Components. Compo-nents of viral DNA synthesis were analyzed in 5-30% neu-tral sucrose gradients (containing 0.01 M Tris, 0.1 M NaCl,0.001 M EDTA, 0.1% Sarkosyl, pH 8.0) and 10-30% alkalinesucrose gradients (containing 0.3 M NaOH, 0.7 M NaCl,0.001 M EDTA, 0.1% Sarkosyl) centrifuged in an SW41rotor at 100 for 8 hr at 40,000 rpm and 12 hr at 40,000 rpm,respectively. Similar gradients were used to recover prepara-tive amounts of specific DNA components. Gradient frac-tionation and assay of radioactivity were described previous-ly (11).The fraction of DNA molecules that contained self-com-

plementary regions was assayed by hydroxyapatite chroma-tography (15), and benzoylated-naphtholated DEAE(BND)-cellulose chromatography (16) was used to isolate thefraction of rapidly reassociating DNA that was completelyor nearly completely double-stranded. Isopycnic banding ofDNA in ethidium bromide-cesium chloride was performedas reported by Bauer and Vinograd (17).

SI Nuclease Digestion of DNA. The single-strand specif-ic nuclease, Si, was purified from crude Aspergillus oryzaepowder (Sigma Chemicals) (18). Reaction mixtures (0.2 ml)

742

Dow

nloa

ded

by g

uest

on

Dec

embe

r 4,

202

0

Proc. Nat. Acad. Sci. USA 73 (1976) 743

10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50FRACTION NUMBER

FIG. 1. Pulse-chase kinetics of AAV replicating DNA mole-cules. Neutral and alkaline sucrose gradient analyses of Hirt ex-tracts of samples obtained after a 2-min pulse (A and D); a 2-minpulse and a 30-min chase (B and E); and a 2-min pulse and 6-hrchase (C and F). Coinfection in this and all subsequent experi-ments was carried out with ts125 as helper. The arrows at 31 S, 21S, and 16 S in neutral gradients indicate positions of cosedimentedmarker Ad5 [14CJDNA, SV40 [14C]DNA I, and SV40 [14C]DNA II.The arrows at 34 S and 16 S in alkaline gradients indicate the posi-tions of marker Ad5 [14C]DNA and SV40 [14C]DNA III.

contained 0.1-0.2 ztg of DNA and 3-15 units of nuclease in30 mM sodium acetate buffer (pH 4.6), 0.5 mM ZnCl2, and75 mM NaCl. After incubation for 15 min at 370, the reac-

tion was terminated by addition of S ,ul of 0.4 M EDTA andthe extent of cleavage determined by sedimentation in neu-

tral and alkaline sucrose gradients. Digestion conditionswere sufficient to completely degrade a 50- to 200-fold ex-

cess of single-stranded DNA and to completely convert an

equivalent amount of simian virus 40 (SV40) DNA I to DNAIII with no detectable single-strand breakage of DNA III.

RESULTSExtraction of AAV DNA. When Hirt extracts of [3H]thy-

midine-labeled viral DNA are sedimented in neutral sucrose

gradients after coinfection with AAV and ts125 or AdS WT(at 39.50), the sedimentation profiles are identical with theexception that adenovirus DNA components (>31 S) are ab-sent when ts125 is the helper (11). With ts125 as helper, la-beled DNA sediments in a multicomponent pattern rangingfrom 10 to 21 S (ref. 11; Figs. 1A, 2, and 4A). Hybridizationanalyses of radioactive virus-specific DNA in both the Hirtsupernatant and pellet fractions after a 1-hr pulse revealedthat about 80% of total labeled AAV (or Ad WT) DNA was

present in the supernatant fraction. Furthermore, in coinfec-tions with AAV and ts125, the 3H-labeled DNA in Hirt ex-

tracts that sedimented from 10 to 21 S in neutral sucrose gra-dients was found to be over 95% AAV-specific both by filterhybridization (11) and by free solution annealing reactionsmonitored by hydroxyapatite chromatography (15).

Because progeny AAV DNA molecules consist of singleplus and minus strands, the analysis of AAV DNA replica-tion requires methods that preserve these specific molecularforms (i.e., duplex molecules must not be artifactually gen-erated from progeny strands during extraction or subsequentsedimentation in neutral sucrose gradients). To be certainthat methods used did not foster annealing of plus andminus progeny strands, we added either alkali-denatured or

annealed (double-stranded) AAV [3H]DNA, obtained frompurified virions, to unlabeled AAV-ts125 coinfected KB cells(16 hr at 39.5°) and then recovered DNA by the modified

O 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70FRACTION NUMBER

FIG. 2. Alkaline sucrose sedimentation of AAV replicatingDNA previously fractionated in a neutral sucrose gradient. DNAfrom the 2-min pulse-labeled preparation analyzed in Fig. 1 (A andD) was sedimented through neutral sucrose (upper left panel) andpools A, B, and C were collected as indicated. Each pool was thencosedimented through alkaline sucrose with a SV40 [14C]DNA IIImarker (16 S). Pool A (A); pool B (B); and pool C (C).

Hirt procedure. In neutral sucrose gradients over 90% of thedenatured AAV [3H]DNA remained single stranded (20 S),whereas over 90% of the annealed AAV [3H]DNA sediment-ed as unit length duplexes (15 S) (data not shown). In addi-tion, an extract was prepared from cells infected for 16 hrwith Ad5 WT alone, pulsed for 10 min with [3H]thymidine,and then chased with thymidine for 6 hr. On alkaline su-

crose sedimentation, the extract contained no radioactiveDNA in molecules shorter than unit length (34 S). Thus, sig-nificant single-strand breakage of DNA does not occur dur-ing the extraction procedure (data not shown).

Identification of Replicating AAV DNA Components.Fig. 1 depicts the results of a pulse-chase experiment afterAAV-ts125 coinfection. Shown are the neutral sucrose sedi-mentation patterns of AAV DNA after a 2-min pulse (A), a

2-min pulse plus a 30-min chase (B), and a 2-min pulse plusa 6-hr chase (C). Panels D, E, and F show the correspondingalkaline sucrose sedimentation patterns. The 2-min pulse-labeled DNA in the 16-21S region of gradient A chases intopart of unit length linear duplex DNA (15 S) (B), which thenchases into single-stranded progeny DNA (20 S) (C). In bothneutral and alkaline gradients the amount of labeled DNAsedimenting more slowly than unit length DNA (10-12 S)appears to be relatively unchanged throughout the chaseand probably represents fragments of duplex DNA. In addi-tion to unit length DNA, the alkaline gradients contain a

peak of faster sedimenting DNA that can be chased into unitlength molecules (16 S in alkaline sucrose). Analysis of thefast sedimenting DNA by electron microscopy (E. Sebring,unpublished results) has failed to reveal covalently closedsingle- or double-stranded circles. Furthermore, radioactivi-ty could not be detected in closed duplex circular moleculeswhen DNA in the 2-min pulse or 6-hr chase extracts was

banded in ethidium bromide-CsCl. Thus, the fast sediment-ing DNA apparently consists of linear molecules that are

31S8 _ I

6

16:4-

2-

ABC

I.I,I| T

A16S

I

I.I AIt

Biochemistry: Straus et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 4,

202

0

744 Biochemistry: Straus et al.

Hirt Supemnatant (4x107 cpm)Dialyzed. Denatured,Cooled, Neutralized,Brought to 0.3M NaClDiluted 20x 8°ND-CelluboseD Chromatography

1M NaCl 20% EthanolEluate Eluate

(2.4 x 107 cpm) la 1.5 x 107 cpm)DS I 62% 38%

Denatured, Cooled 0C,Neutralized, 1 M NaCl

BND-CelluloseChromatography

1M NaCl 20% EthanolFlow Through Eluate

DS 19.5 x 106 cpm (1.4 x 107 cpm)DSI 24% 35%

Diluted in0.3M NaCl

BND-CelluloseChromatography

1M NaClI 2 20% EthanolEluate Eluate

(7.5 x 106 cpm) 0n (1.4 x 106 cpm)DSm 19% 3.5%

Proc. Nat. Acad. Sci. USA 73 (1976)

16

Ss IX; 11

8

ss I

A

FRACTION NUMBER

sS

FIG. 3. Purification of AAV DNA replicating snap-back mole-cules.

longer than progeny strands. The location of long moleculeswith respect to the neutral sucrose sedimentation profileproved to be interesting (Fig. 2). When fractions in the 15Sregion from a preparative neutral sucrose gradient of theDNA after the 2-min pulse were pooled and sedimented inalkaline sucrose, about half of this DNA (unit length duplexDNA in neutral sucrose) sedimented faster than 16S unitlength single strands, with a distinct peak at 21S which istwice unit length size (Fig. 2B). Additionally, at least half ofthe 16-21S DNA from the neutral gradient also sedimentedfaster than unit length strands (Fig. 2A), whereas most of the10-12S DNA sedimented more slowly (Fig. 2C). These dataindicate that (a) there are AAV DNA RI that contain contin-uous linear strands whose lengths exceed that of single-stranded progeny molecules and (b) the long strands may befolded into partially or completely duplex structures (e.g.,hairpin-like molecules). The specific nature of the rapidlylabeled but seemingly "dead end" 10-12S DNA is uncertain.

Concatemeric DNA Molecules. When a Hirt extract ofAAV-ts125 coinfected KB cells, pulse-labeled for 10 min,was denatured with 0.1 M NaOH, chilled to 00, neutralized,and chromatographed on hydroxyapatite, 55% of the labeledDNA eluted as double-stranded DNA. Thus, a large portionof the pulse-labeled AAV DNA contains self-complementaryregions that rapidly reassociate to form duplex DNA, a resultconsistent with postulation (b) above. To further character-ize these "snap-back" molecules, BND-cellulose, which se-lectively retains DNA with single-stranded regions, was usedto isolate the fraction of rapidly reassociating DNA that iscompletely or almost completely double-stranded. ExtractedDNA that had been labeled for 1 hr with [3H]thymidine wasdenatured with 0.1 M NaOH. After neutralization, double-stranded DNA was isolated and the procedure repeated asecond time. Approximately 19% of the starting radioactivi-ty was recovered in the final double-stranded fraction (DS

20 40 60

40

30

20

10 _

25

20

15

10

FIG. 4. Sucrose gradient analysis of rapidly reassociatingDNA. Hirt supernatant AAV replicating DNA and the fraction ofrapidly reassociating DNA (DS III), isolated as depicted in Fig. 3,were each analyzed in neutral and alkaline gradients. Neutral su-crose sedimentation of (A) Hirt supernatant DNA and (B) DS IIIDNA; alkaline sucrose sedimentation of (C) Hirt supernatantDNA and (D) DS III DNA. Cosedimented marker SV40 [14CJDNAradioactivity is plotted with a dashed line. Preparative neutral su-crose gradients of DS III DNA were also run and divided into twopools (1 and 2) as shown in panel (B).

III) (Fig. 3). Fig. 4 shows neutral and alkaline sucrose sedi-mentation patterns of AAV DNA in the Hirt supernatant (Aand C) and the DS III BND-cellulose fraction (B and D).The DS III alkaline gradient (D) demonstrates a marked en-richment for DNA sedimenting faster than unit length witha pronounced peak sedimenting as molecules that weretwice unit length (21 S). To define more clearly the natureof these long molecules, we pooled fractions from the neu-tral sucrose gradient of rapidly reassociated DNA, as indicat-ed in Fig. 4B, and each pool was sedimented in alkaline su-crose, as shown in Fig. 5. Double-stranded DNA that sedi-mented ahead of unit length duplexes in neutral sucrose(pool 1) mostly consisted of strands greater than unit length(Fig. 5A), and about two-thirds of molecules that sediment-ed in the region of unit length duplexes (pool 2) were alsolonger than unit length strands (Fig. 5B). Approximately40% of this latter DNA sedimented as molecules that were

16 40 -

16S12 A - 30

8-20

4 i.; ~~~~~~~~~~~~~~-10'I

0 ~ ~0

FRACTION NUMBER20 40 60

FIG. 5. Alkaline sucrose sedimentation of DS III DNA pools 1and 2 which were fractionated as shown in Fig. 4B. Pool 1 (A); pool2 (B). Cosedimented marker SV40 [14C]DNA III radioactivity isplotted with a dashed line. Fractions corresponding to a tetramer(27 S) and peak fractions corresponding to a dimer (21 S) were col-lected from similarly run preparative gradients and designatedpool I and pool II, as shown in (A) and (B).

B

40 16 1 2r - -In21S 15S

30 12 IDe

D i 16S

8

21S tO0.,

4-

2~~~~~~~~~~~~~~~~~~~~~~~~~~6 Qq

GSJ '

21S12 j 15S

9 iI

:1.8 Al -

4 -

0

I:

0-

Uom

I)

Dow

nloa

ded

by g

uest

on

Dec

embe

r 4,

202

0

Proc. Nat. Acad. Sci. USA 73 (1976) 745

20 40 60 20 40 60FRACTION NUMBER

FIG. 6. Neutral and alkaline sucrose sedimentation analyses oftetramer and dimer length AAV DNA strands (pools I and II, Fig.5). (A) Neutral and (B) alkaline sedimentation of tetramer pool;(C) neutral and (D) alkaline sedimentation of dimer pool. Cosedi-mented marker SV40 [14C]DNA radioactivity is plotted with a

dashed line.

twice unit length. Finally, fractions from the first gradientcorresponding to a tetramer (27 S) were combined (Fig. 5A,pool I) as well as peak fractions from the second gradientcorresponding to a dimer (Fig. SB, pool II) and each poolwas resedimented in neutral and alkaline sucrose (Fig. 6.).Comparison of the respective neutral and alkaline sedimen-tation patterns indicates that on neutralization the denaturedmolecules can fold into duplex structures of approximatelyhalf single strand length. The most likely structure of mole-cules with these sedimentation properties would be linearstrands composed of covalently end-to-end linked plus andminus strands of AAV DNA. The bimodal sedimentationpattern observed in the neutral gradient of pool I DNA (Fig.6A) may have arisen because of different folding configura-tions possible with a tetramer composed of alternating plusand minus strands (i.e., the 18S component would representa linear hairpin twice unit length, whereas the 21S compo-

nent might represent a branched duplex form).Further evidence that replicating DNA can exist as a hair-

pin structure was obtained by Si nuclease digestion of therapidly reassociated unit length duplexes. A hairpin mole-cule would necessarily have several unpaired bases in thehairpin loop, and Si nuclease, which is highly specific forsingle-stranded DNA, would be expected to cleave this loop.Denaturation of treated molecules should then give rise tounit length single strands. Fig. 7 shows alkaline sucrose gra-

dients of rapidly reassociated unit length duplexes beforeand after Si nuclease treatment. Clearly, molecules thatsediment in alkaline sucrose as 21S dimers are cleaved bySI nuclease to molecules that sediment as 16S unit lengthstrands.

DISCUSSIONLittle is known concerning the actual DNA synthetic mech-anism of any parvovirus, because the structures of replicat-ing DNA molecules have not been sufficiently character-ized. In the case of minute virus of mice and Kilham ratvirus (both nondefective parvoviruses) there is evidence fora linear, duplex replicating intermediate (19, 20). For min-

10 20 30 40 50 60 70FRACTION NUMBER

50t 2X 1X AFTER S. 25

40 20

30k-11

20 h

10

C 00 10 20 30 40 50 60 70

FRACTION NUMBER

FIG. 7. Treatment of dimer length strands of rapidly reasso-ciating AAV replicating DNA with S1 nuclease. Dimer strands ob-tained as in Fig. 6D were further purified by sedimentationthrough neutral sucrose. The unit length duplex peak was collect-ed and dialyzed into 0.01 M Tris-0.01 M NaCl, pH 7.5. Portions ofthis DNA were treated with 2, 5, or 10 til of S1 nuclease (1600units/ml), and the samples as well as untreated DNA were ana-lyzed in alkaline sucrose. Only the sample treated with 5 1A of S1nuclease is shown, since all three enzyme concentrations gave iden-tical sedimentation patterns. Cosedimented marker SV40[14C]DNA radioactivity is plotted with a dashed line.

ute virus of mice, some of these molecules were shown to becapable of spontaneous, intramolecular renaturation, but thespecific structural feature responsible for renaturation wasnot defined. This observation, however, prompted the spec-ulation that minute virus of mice DNA synthesis might bemediated by a self-primed hairpin intermediate (19).Our results indicate that AAV DNA RI contain continuous

linear strands whose lengths may reach or possibly exceedfour times the length of progeny strands, and that these longmolecules appear to be concatemers composed of alternatingplus and minus strands. Of the population of concatemerscapable of self-annealing into completely or almost com-pletely double-stranded molecules, the relatively largest andmost discrete fraction consists of dimers that presumablyexist as linear hairpin structures in vivo (Fig. 2B). LongerDNA concatemers might also possess this same conformationduring DNA replication. However, other structures are pos-sible, and branched molecules can be found when prepara-tions of RI isolated by BND-cellulose chromatography areexamined in the electron microscope (E. Sebring, unpub-lished results). Although we are not yet able to exclude arolling circle mechanism for AAV DNA synthesis, electronmicroscopic studies of AAV DNA RI do not support this pos-sibility.The observations that (a) the immediate precursor of

progeny strands is a unit length duplex (Fig. 1), (b) a unitlength hairpin is a major DNA intermediate (Figs. 2 and 4),and (c) RNA sequences do not appear to be covalentlyjoined to replicating DNA molecules (stability of concatem-ers in alkali; S. Straus, unpublished results) are compatiblewith a replication scheme in which DNA synthesis is initiat-ed by a self-priming mechanism. A model for the formationof a unit length hairpin molecules as an intermediate inAAV DNA replication is shown in Fig. 8A. The self-comple-mentary terminal segments of AAV DNA strands (5) arepostulated to contain a palindromic sequence (7) whichcould fold back on itself to form a short duplex region. The3' end of one or both strands could then serve as a primer forsynthesis of the complementary strand, generating a hairpin.Based on this mechanism, one of several possible models forsynthesis of single-stranded progeny molecules is depicted inFig. 8B and C. Fig. 8B shows how the hairpin molecule (Fig.8A) might be converted to a unit length duplex with com-plete ends according to steps recently proposed by Cavalier-

Biochemistry: Straus et al.

x

Q0

Z_6

6

O.

8o

x0

un

8fII

Dow

nloa

ded

by g

uest

on

Dec

embe

r 4,

202

0

Proc. Nat. Acad. Sci. USA 73 (1976)

bI ATBe .------AT) *

A 5.5ABC*BA ABC*BA'f 7 5' -i_5- jV@- 7 ~~~~~ABAB-

3' AT 3' ATC*BA5 3 ATC*BA#ABzzzzCt. ____ ABC*BA'

(4.)

5''3' 5' 5~~~~~~~~~~~~~~' 5

OR 5' 5'~~~~~~5

D ,, 3 _ 5'

D 3

5- ---3-_+ F

__ _ _-- --3

FIG. 8. Model for replication of AAV DNA. In (A) and (B) let-ters represent terminal sequences equivalent to a total of 100-200nucleotides. Sequence complementarity is indicated by corre-sponding primed and unprimed letters. The letter C* represents asequence that contains unpaired bases in the hairpin loop. To pro-vide identical molecular ends, this sequence would have to be apalindrome (as shown).

Smith (21). After a nick (arrow) near the closed end of thehairpin, the 5-ended segment, containing a full complementof terminal sequences, becomes available as a template forcompleting the other strand by 5'-3' synthesis. A plus orminus strand could then be generated from the resulting du-plex intermediate by displacement synthesis, which againinvolves self-priming (Fig. 8C). Initiation of displacementsynthesis by either strand in the duplex template may relateto the presence of identical molecular ends, a consequenceof the inverted repetition (5). The unit length hairpins pro-duced along with single strands could be processed (Fig. 8B)to provide more templates for displacement synthesis. Con-sistent with the expectation that a connecting loop wouldoccur at either end of the hairpin molecule (Fig. 8C), wehave recently found that when unit length hairpins arecleaved by endonuclease R.EcoRI (22), and then denatured,approximately half of both right- and left-end fragmentsundergoes rapid renaturation (unpublished results). Al-though hairpin molecules have not been identified as inter-mediates in adenovirus DNA replication, it is notable thatthese AAV helpers also utilize displacement synthesis to pro-duce their own DNA (23).

Evidence that most of the radioactivity in unit lengthhairpins moves into unit length single strands during a chaseindicates that these hairpin molecules are directly convertedto templates for displacement synthesis. An alternative path-way may account for plus-minus concatemers of greaterthan twice unit length (Fig. 8D). The 3' end of a unit lengthhairpin would displace the 5' end of its complementarystrand, priming the synthesis of a hairpin twice unit length

(or longer).t Such molecules might then be cleaved by astaggered cut (arrows) as proposed for T7 DNA (24), and the3' ends regenerated by 5'-3' synthesis giving a unit lengthlinear duplex (with complete ends) and hairpin.

Processing of concatemers to unit length single strands(Fig. 1) implies the presence of a sequence-specific endonu-clease(s) which might act as suggested in Fig. 8B and D.This activity could reside in an AAV, adenovirus, or cellularprotein and is specially notable because a restriction-like en-donuclease has not yet been isolated from an animal cell.

1. Rose, J. A. (1974) in Comprehensive Virology, eds. Fraenkel-Conrat, H. & Wagner, R. (Plenum Press, New York), Vol. 3,pp. 1-61.

2. Carter, B. J., Koczot, F. J., Garrison, J., Dolin, R. & Rose, J. A.(1973) Nature New Biol. 244,71-73.

3. Rose, J. A., Berns, K. I., Hoggan, M. D. & Koezot, F. J. (1969)Proc. Nat. Acad. Sci. USA 64,863-869.

4. Berns, K. I. & Rose, J. A. (1970) J. Virol. 5, 693-699.5. Koczot, F. J., Carter, B. J., Garon, C. F. & Rose, J. A. (1973)

Proc. Nat. Acad. Sci. USA 70,215-219.6. Berns, K. I. & Kelly, T. J., Jr. (1974) J. Mol. Biol. 82, 267-271.7. Gerry, H. W., Kelly, T. J., Jr. & Berns, K. I. (1973) J. Mol.

Biol. 79,207-225.8. Garon, C. F., Berry, K. W. & Rose, J. A. (1972) Proc. Nat.

Acad. Sci. USA 69,2391-2395.9. Wolfson, J. & Dressler, D. (1972) Proc. Nat. Acad. Sci. USA

69,3054-3057.10. Garon, C. F., Berry, K. W. & Rose, J. A. (1975) Proc. Nat.

Acad. Sci. USA 72, 3039-3043.11. Straus, S. E., Ginsberg, H. S. & Rose, J. A. (1976) J. Virol. 17,

140-148.12. Ensinger, M. J. & Ginsberg, H. S. (1972) J. Virol. 10, 328-339.13. Lavelle, G., Patch, C., Khoury, G. & Rose, J. A. (1975) J.

Virol. 16, 775-782.14. Rose, J. A. & Koczot, F. J. (1972) J. Virol. 10, 1-8.15. Thoren, M. M., Sebring, E. D. & Salzman, N. P. (1972) J.

Virol. 10, 462-468.16. Sedat, J. W., Kelly, R. B. & Sinsheimer, R. L. (1967) J. Mol.

Biol. 26,537-540.17. Bauer, W. & Vinograd, J. (1968) J. Mol. Biol. 33, 141-171.18. Vogt, V. (1973) Eur. J. Biochem. 33, 192-200.19. Tattersall, P., Crawford, L. V. & Shatkin, A. J. (1973) J. Virol.

12, 1446-1456.20. Salzman, L. A. & White, W. L. (1973) J. Virol. 11, 299-305.21. Cavalier-Smith, T. (1974) Nature 250,467-470.22. Carter, B. J. & Khoury, G. (1975) Virology 63,523-538.23. Sussenbach, J. S., van der Vliet, P. C., Ellens, D. J. & Jansz, H.

S. (1972) Nature New Biol. 239,47-49.24. Watson, J. D. (1972) Nature New Biol. 239, 197-201.

t AAV DNA synthesis would thus be moving back and forth in aboustrophedonic manner, i.e., like the method of ancient writingin which lines were written from right to left and from left toright.

746 Biochemistry: Straus et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 4,

202

0